8/10/2019 USC NancyDeng MolecularDockingStudies
1/58
Structure-based Drug Design:Structure-based Drug Design:Molecular Docking StudiesMolecular Docking Studies
Jinxia (Nancy) Deng, PhDJinxia (Nancy) Deng, PhDSept 25, 2007Sept 25, 2007
8/10/2019 USC NancyDeng MolecularDockingStudies
2/58
OutlineOutline
Review of Structure Based DrugReview of Structure Based Drug
DesignDesign
Molecular dockingMolecular docking
Docking case studyDocking case study
8/10/2019 USC NancyDeng MolecularDockingStudies
3/58
http://www-personal.engin.umich.edu/~wildd/mfg501/lecture.ppt
Identify disease
Isolate protein
involved in
disease (2-5 years)
Find a drug effective
against disease protein
(2-5 years)
Preclinical testing
(1-3 years)
Formulation
Human clinical trials
(2-10 years)
Scale-up
FDA approval
(2-3 years)
F i l e
I N D
F i l e
N D A
http://images.google.com/imgres?imgurl=www.elements.nb.ca/theme/health/patty/sick.jpg&imgrefurl=http://www.elements.nb.ca/theme/health/theme.htm&h=128&w=75&prev=/images%3Fq%3Dsick%2Bclipart%26svnum%3D10%26hl%3Den%26lr%3D%26ie%3DUTF-8%26safe%3Doff
8/10/2019 USC NancyDeng MolecularDockingStudies
4/58
Database searchingDatabase searching
3D structure searching3D structure searching pharmacophore modelspharmacophore models exclusion zones to represent receptorexclusion zones to represent receptor
DockingDocking prediction of the structure and binding free energyprediction of the structure and binding free energy
of a ligand-receptor complex given only theof a ligand-receptor complex given only thestructures of the free ligand and receptorstructures of the free ligand and receptor
examine binding model of known ligands toexamine binding model of known ligands tosuggest modificationsuggest modification
screen databases of 3D structure to find novelscreen databases of 3D structure to find novelligandsligands
8/10/2019 USC NancyDeng MolecularDockingStudies
5/58
It includesIt includes ligand-protein docking, one branch ofligand-protein docking, one branch of
rational drug design to predict whichrational drug design to predict which
molecules can displace others frommolecules can displace others fromthe surface of a protein, or changethe surface of a protein, or change
protein function.protein function.
Protein-protein interactionProtein-protein interaction
Protein – DNA interactionProtein – DNA interaction
8/10/2019 USC NancyDeng MolecularDockingStudies
6/58
Ligand docking(lock & key model)
8/10/2019 USC NancyDeng MolecularDockingStudies
7/58
8/10/2019 USC NancyDeng MolecularDockingStudies
8/58
Structure
determination
Design new
inhibitors
Targetstructure Small
molecule
database
8/10/2019 USC NancyDeng MolecularDockingStudies
9/58
•Ligand-protein docking:
• Docking of a small molecule (the ligand) on
a large molecule (the protein, so called-receptor)
•Protein-protein docking:
• Usually the docking site is a more "planar" surface than inthe ligand-protein docking
•Protein-DNA simulation*
Nucleic Acids Research 2006, Vol34, No. 11 3317-3325
8/10/2019 USC NancyDeng MolecularDockingStudies
10/58
Evaluating a structureEvaluating a structure
X-ray crystallographyX-ray crystallography Most commonMost common Must have better than 2.5Must have better than 2.5ÅÅ
resolutionresolution
NMRNMR May be possible to determineMay be possible to determine
dynamicsdynamics
Homology modelingHomology modeling When no experimental structureWhen no experimental structure
is availableis available Uses a known structure as aUses a known structure as a
template for the unknowntemplate for the unknown
8/10/2019 USC NancyDeng MolecularDockingStudies
11/58
Growth of the PDBGrowth of the PDB
0
5000
10000
15000
20000
25000
30000
1977 1982 1987 1992 1997 2002
Year
N u m b e r o
f E n t r i e s
http://www.rcsb.org
8/10/2019 USC NancyDeng MolecularDockingStudies
12/58
Identification of target siteIdentification of target site
Ideally a pocket or protuberance withIdeally a pocket or protuberance with Hydrogen bond donors and acceptorsHydrogen bond donors and acceptors
Hydrophobic characteristicsHydrophobic characteristics
Variety of sizes of molecular surfacesVariety of sizes of molecular surfaces Can be an active site, an assembly site,Can be an active site, an assembly site,
or a communication siteor a communication site
Co-crystallized protein-ligand complexesCo-crystallized protein-ligand complexescan be invaluable for determining targetcan be invaluable for determining target
sitessites
8/10/2019 USC NancyDeng MolecularDockingStudies
13/58
Docking algorithmsDocking algorithms
Molecular flexibilityMolecular flexibility both ligand and protein rigidboth ligand and protein rigid flexible ligand and rigid proteinflexible ligand and rigid protein both ligand and protein flexibleboth ligand and protein flexible
search algorithmsearch algorithm use to explore optimal positions of the ligand withinuse to explore optimal positions of the ligand within
the active sitethe active site
scoring functionscoring function value should correspond to preferred binding modevalue should correspond to preferred binding mode
efficiency very important for database searchingefficiency very important for database searching
8/10/2019 USC NancyDeng MolecularDockingStudies
14/58
TODAY’S FOCUSTODAY’S FOCUS
eHITSeHITS:: EElectroniclectronic HiHighgh
TThroughputhroughput SScreeningcreening
((SimBioSys Inc.)SimBioSys Inc.)
GOLDGOLD:: GGeneticenetic OOptimizationptimizationforfor LLigandigand DDocking (ocking (CCDC CCDC ))
http://www.ccdc.cam.ac.uk/prods/gold/index.htmlhttp://www.ccdc.cam.ac.uk/prods/gold/index.htmlhttp://www.ccdc.cam.ac.uk/prods/gold/index.html
8/10/2019 USC NancyDeng MolecularDockingStudies
15/58
eHITS: Electronic High ThroughputeHITS: Electronic High Throughput
ScreeningScreening (SimBioSys Inc.)(SimBioSys Inc.)
Fast, felxibibleFast, felxibible
docking of whole anddocking of whole andpartial structures topartial structures totarget receptorstarget receptors
Designed for libraryDesigned for libraryscreeningscreening
8/10/2019 USC NancyDeng MolecularDockingStudies
16/58
eHiTS - SearcheHiTS - Search
❑ Ligand is divided intoLigand is divided into❑ rigid fragments andrigid fragments and
❑ connecting flexible chainsconnecting flexible chains
❑ All rigid fragments areAll rigid fragments are
docked independentlydocked independently
❑ Graph matchingGraph matching
❑ Flexible chain fittingFlexible chain fitting
❑ Local energy minimisationLocal energy minimisation
8/10/2019 USC NancyDeng MolecularDockingStudies
17/58
8/10/2019 USC NancyDeng MolecularDockingStudies
18/58
There are many docking programsThere are many docking programs eHITS is a fully systematic and exhaustiveeHITS is a fully systematic and exhaustive
docking programdocking program
19 F B d HiTSF t B d HiTS
8/10/2019 USC NancyDeng MolecularDockingStudies
19/58
1919.. Fragment Based eHiTSFragment Based eHiTS
AlgorithmAlgorithm
Ligands are divided into rigidLigands are divided into rigidfragments and flexible connectingfragments and flexible connectingchainschains
Rigid Dock: Each fragment isRigid Dock: Each fragment isdocked INDEPENDENTLYdocked INDEPENDENTLY
everywhere in the receptoreverywhere in the receptor Pose Match: A fast graphPose Match: A fast graph
matching algorithm finds allmatching algorithm finds allmatching solutions to reconstructmatching solutions to reconstruct
the original moleculethe original molecule Local Energy Optimization:Local Energy Optimization:
structure is optimized within thestructure is optimized within thereceptorreceptor
Ranking: structures are rankedRanking: structures are rankedbased on scoring functionbased on scoring function
N
O
N
O
N
O
N
O
HN
N
N
OH2
HN
N
HN
NHN
N
HN
N
HN
N
NH
N
HN
N
H2H
2
H2
H2
H2
H2
H2
Reconnected LigandPose:
HN
N
N
O
H2
8/10/2019 USC NancyDeng MolecularDockingStudies
20/58
2020.. Docking Fragments - RigiDockDocking Fragments - RigiDock
Rigid fragment docking based onRigid fragment docking based onChemical feature mapped polyhedraChemical feature mapped polyhedra
Polyhedron shrink-wrapped ontoPolyhedron shrink-wrapped onto
molecular surface (Connolly)molecular surface (Connolly)
Chemical feature flags on verticesChemical feature flags on vertices
Analogue cavity representationAnalogue cavity representation
Rapid mapping of ligand and cavityRapid mapping of ligand and cavity
polyhedrapolyhedra
8/10/2019 USC NancyDeng MolecularDockingStudies
21/58
2121.. Fragment Based DockingFragment Based Docking
8/10/2019 USC NancyDeng MolecularDockingStudies
22/58
2222.. Docking Fragment 1Docking Fragment 1
The Fragment is docked
everywhere in the receptorsite, and clustered to givea set number of poses
NOTE: In the illustrations shown, only 250 poses are displayed,default parameters of eHiTS uses 60 for ea!h fra"ment
8/10/2019 USC NancyDeng MolecularDockingStudies
23/58
2323.. Docking Fragment 2Docking Fragment 2
8/10/2019 USC NancyDeng MolecularDockingStudies
24/58
2424.. Docking Fragment 3Docking Fragment 3
8/10/2019 USC NancyDeng MolecularDockingStudies
25/58
2525.. Docking Fragment 4Docking Fragment 4
8/10/2019 USC NancyDeng MolecularDockingStudies
26/58
2626.. Close but differentClose but different
Polar CharacterFragment 4: Alcohol Oxygen
Fragment 2: Ether Oxygen
8/10/2019 USC NancyDeng MolecularDockingStudies
27/58
F 1 F 2 F 3 F 4
8/10/2019 USC NancyDeng MolecularDockingStudies
28/58
Top Ranked PoseRMSD: 0.699
Score: -8.507
Frag 1 Frag 2 Frag 3 Frag 4
8/10/2019 USC NancyDeng MolecularDockingStudies
29/58
2929.. Summary of eHiTS FeaturesSummary of eHiTS Features
Fragment based docking engineFragment based docking engine
Speed-up due to database storageSpeed-up due to database storage
Statistically derived empirical scoringStatistically derived empirical scoringfunctionfunction
User-trainable scoring function (FamilyUser-trainable scoring function (Family
based)based) Automatic handling of protonation stateAutomatic handling of protonation state
8/10/2019 USC NancyDeng MolecularDockingStudies
30/58
DHFR Case StudyDHFR Case Study
8/10/2019 USC NancyDeng MolecularDockingStudies
31/58
Experiment ObjectivesExperiment Objectives
❑ Show ability to reproduce crystalShow ability to reproduce crystal
structuresstructures
❑ Show that eHiTS can select active ligandsShow that eHiTS can select active ligandsof human DHFR from a drug databaseof human DHFR from a drug database
❑ Illustrate the ease of use of eHiTSIllustrate the ease of use of eHiTS
❑ No pdb preparation, no ligandNo pdb preparation, no ligand
preparationpreparation
❑
Show eHiTS can be used in HTSShow eHiTS can be used in HTS
8/10/2019 USC NancyDeng MolecularDockingStudies
32/58
Dihydrofolate ReductaseDihydrofolate Reductase
(DHFR)(DHFR)❑ Plays an essential role in thePlays an essential role in the
building of DNAbuilding of DNA
❑
““ juggles” two molecules in juggles” two molecules inthis reactionthis reaction
❑ Folate (purple) and NADPHFolate (purple) and NADPH
(green)(green)❑ The first enzyme targeted forThe first enzyme targeted for
cancer chemotherapycancer chemotherapy Oct. 2002 PDB Molecule of the Month:http://www.rcsb.org/pdb/molecules/pdb34_3.html
8/10/2019 USC NancyDeng MolecularDockingStudies
33/58
DHFR – Binding siteDHFR – Binding site
❑ The drug methotrexate is designed toThe drug methotrexate is designed to
mimic folate, blocking the enzyme's actionmimic folate, blocking the enzyme's action
❑
Note the interaction between folate andNote the interaction between folate andNADPH, this is essential for the enzyme'sNADPH, this is essential for the enzyme's
functionfunction
8/10/2019 USC NancyDeng MolecularDockingStudies
34/58
““Actives” SelectionActives” Selection❑
Searched for DHFR complexes in the PDBSearched for DHFR complexes in the PDB❑ Obtained 88 complexes, all sourcesObtained 88 complexes, all sources
❑ Upon quick visual inspection, eliminated 17Upon quick visual inspection, eliminated 17
complexescomplexes❑ Contained no ligand in the binding siteContained no ligand in the binding site
❑ Contained multiple ligands in the binding siteContained multiple ligands in the binding site
❑ Selected 71 DHFR complexes for studySelected 71 DHFR complexes for study
❑ Including 19 human complexesIncluding 19 human complexes
8/10/2019 USC NancyDeng MolecularDockingStudies
35/58
19 Human DHFR complex19 Human DHFR complex
Com plex Ligand Form ula # at om s CoF Form ula 1dhf FOL 2(C19 H17 N7 O6 --) 49
1dlr MXA C17 H19 N5 O2 43 NDP C21 H27 N7 O17 P 1dls MTX C20 H22 N8 O5 55 NDP C21 H27 N7 O17 P
1drf FOL C19 H17 N7 O6 -- 49 1hfp MOT C20 H22 N6 O6 54 NAP C21 H28 N7 O17 P
1hfq MOT C20 H22 N6 O6 54 NAP C21 H28 N7 O17 P 1hfr MOT C20 H22 N6 O6 54 NAP C21 H28 N7 O17 P 1km s LIH C18 H17 N7 42 NDP C21 H30 N7 O17 P 1km v LII C18 H19 N5 O2 44 NDP C21 H30 N7 O17 P
1m vs DTM C18 H22 N6 O3 49 1m vt DTM C18 H22 N6 O3 49
1ohj COP C27 H27 N9 O6 69 NDP C21 H30 N7 O17 P 1ohk COP C27 H27 N9 O6 69 NDP C21 H30 N7 O17 P 1pd8 CO4 C19 H24 N6 O3 52 NDP C21 H30 N7 O17 P
1pd9 CO4 C19 H24 N6 O3 52 1s3u TQD C19 H39 N5 O3 66 1s3v TQD C19 H39 N5 O3 66
1s3w TQT C17 H33 N5 55 NAP C21 H28 N7 O17 P 2dhf DZF 2(C20 H18 N6 O6 --) 50
8/10/2019 USC NancyDeng MolecularDockingStudies
36/58
ValidationValidation
❑ Each DHFR ligand was removed from theEach DHFR ligand was removed from the
protein and docked back into its binding siteprotein and docked back into its binding site
❑ EHiTS was allowed to do this split automaticallyEHiTS was allowed to do this split automatically
❑ Results were then judged by evaluating theResults were then judged by evaluating the
RMSD between the crystal structure bindingRMSD between the crystal structure binding
position and the computed docking poseposition and the computed docking pose
❑ Standard (default) parameters for eHiTS wereStandard (default) parameters for eHiTS were
used in all the runsused in all the runs
8/10/2019 USC NancyDeng MolecularDockingStudies
37/58
Validation – All SourcesValidation – All Sources
Top-ranked Closest
< 0.5 3.23% 17.74%< 1.0 22.58% 51.61%
< 1.5 59.68% 69.35%
< 2.0 67.74% 85.48%< 2.5 83.87% 91.94%
< 3.0 88.71% 91.94%Ave RMSD 1.94 1.41
8/10/2019 USC NancyDeng MolecularDockingStudies
38/58
Validation - HumanValidation - Human
8/10/2019 USC NancyDeng MolecularDockingStudies
39/58
Results – The GoodResults – The Good
1dyi Complex – x-ray ligand in white
Top-Rank, -139.60.85 RMS
Closest, -105.770.76 RMS
8/10/2019 USC NancyDeng MolecularDockingStudies
40/58
Results – The BadResults – The Bad
1ly4 Complex – x-ray ligand in white
Top-Rank, -55.112.23 RMS
Closest, -50.320.89 RMS
8/10/2019 USC NancyDeng MolecularDockingStudies
41/58
Results – The UglyResults – The Ugly
Top-Rank, -7.874.90 RMS
Closest, 43.224.38 RMS
1rc4 Complex – x-ray ligand in white
8/10/2019 USC NancyDeng MolecularDockingStudies
42/58
Validation - SummaryValidation - Summary
❑ EHiTS was able to reproduce accuratelyEHiTS was able to reproduce accurately
(RMS < 2.0) the crystal structure position of(RMS < 2.0) the crystal structure position of
DHFR ligands 85% of the timeDHFR ligands 85% of the time
❑ 67% of the time, eHiTS' highest ranking67% of the time, eHiTS' highest ranking
(best scoring) pose had a RMS < 2.0(best scoring) pose had a RMS < 2.0
❑
This number improves for Human DHFRThis number improves for Human DHFRligands, 74%ligands, 74%
❑ This shows that eHiTS is able to predictThis shows that eHiTS is able to predict
docking poses for DHFR ligandsdocking poses for DHFR ligands
GOLDGOLD:: GGeneticenetic OOptimization forptimization for
http://www.ccdc.cam.ac.uk/prods/gold/index.htmlhttp://www.ccdc.cam.ac.uk/prods/gold/index.htmlhttp://www.ccdc.cam.ac.uk/prods/gold/index.html
8/10/2019 USC NancyDeng MolecularDockingStudies
43/58
calculating docking modes of smallcalculating docking modes of small
molecules into protein binding sitesmolecules into protein binding sites genetic algorithm for protein-ligandgenetic algorithm for protein-ligand
dockingdocking
full ligand and partial protein flexibilityfull ligand and partial protein flexibility energy functions partly based onenergy functions partly based onconformational and non-bonded contactconformational and non-bonded contactinformation from the CSDinformation from the CSD
choice of scoring functions: GoldScore,choice of scoring functions: GoldScore,ChemScore and User defined scoreChemScore and User defined score virtual library screeningvirtual library screening
GOLDGOLD:: GGeneticenetic OOptimization forptimization for
LLigandigand DDocking(ocking( CCDC CCDC ))
GOLD t iGOLD t i
http://www.ccdc.cam.ac.uk/prods/gold/index.htmlhttp://www.ccdc.cam.ac.uk/prods/gold/index.htmlhttp://www.ccdc.cam.ac.uk/prods/gold/index.html
8/10/2019 USC NancyDeng MolecularDockingStudies
44/58
GOLD contains:GOLD contains:
A scoring function:A scoring function: GOLD Fitness =SGOLD Fitness =Shb-ext +hb-ext +SSvdw-exvdw-ex+S+Shb-inthb-int+S+Svdw-intvdw-int
Searching mechanism to explore bindingSearching mechanism to explore binding
mode: Genetic algorithm.mode: Genetic algorithm.
Docking of the ligandin 1BL7, a MAPkinase P38 complex.The GOLD dockedsolution (red) iscompared with thatobserved in the PDB(coloured by
element). RMSD =0.5 Angstroms.
Docking of the ligand in1JAP, a matrixmetalloproteasecomplex. The GOLDdocked solution (red) iscompared with thatobserved in the PDB(coloured by element)RMSD = 0.8 Angstroms
Genetic Algorithm: selection crossover and mutation
8/10/2019 USC NancyDeng MolecularDockingStudies
45/58
Genetic Algorithm: selection , crossover , and mutationoperators to find the optimal solution
Solutions are encoded as genome, i.e.Solutions are encoded as genome, i.e.chromsomechromsome Chromosome A:Chromosome A:
101100101100101011100101101100101100101011100101 Chromosome B:Chromosome B:
111111100000110000011111111111100000110000011111
8/10/2019 USC NancyDeng MolecularDockingStudies
46/58
Reproduction: via crossover, which picksReproduction: via crossover, which picks
randomly from parent genome to producerandomly from parent genome to producenew generation.new generation.
8/10/2019 USC NancyDeng MolecularDockingStudies
47/58
Mutation: chromosome can undergoMutation: chromosome can undergo
spontaneous changes (with smallspontaneous changes (with smallprobability) to produce new generations toprobability) to produce new generations to
sampling different part of the solutionsampling different part of the solution
space.space.
AA fitness function fitness function (or(or objective function objective function ) is) is
used to evaluate each solution.used to evaluate each solution.
8/10/2019 USC NancyDeng MolecularDockingStudies
48/58
Docking applicationsDocking applications
Binding mode predictionBinding mode prediction
Novel lead design or optimizationNovel lead design or optimization
Database screening to identify hitsDatabase screening to identify hits
Design novel leads from
8/10/2019 USC NancyDeng MolecularDockingStudies
49/58
Design novel leads fromfragment
target
Docking
fragment
LinkagePerfect ligand
8/10/2019 USC NancyDeng MolecularDockingStudies
50/58
8/10/2019 USC NancyDeng MolecularDockingStudies
51/58
Challenging: receptor flexibilityChallenging: receptor flexibility
Ligand binding to the receptor induces a range ofLigand binding to the receptor induces a range of
conformational changes:conformational changes: Local rearrangement of side-chainsLocal rearrangement of side-chains
Hinge motions of domainHinge motions of domain Two problemsTwo problems
Predicting conformational changePredicting conformational change
sampling the possible changed conformation duringsampling the possible changed conformation during
docking because too many degree of freedom available todocking because too many degree of freedom available toa binding site and number of binding site conformationsa binding site and number of binding site conformations
grows exponentially with them.grows exponentially with them.
8/10/2019 USC NancyDeng MolecularDockingStudies
52/58
Virtual Screening / DockingVirtual Screening / Docking
eHiTSDOCK
FlexX
FlexE
SLIDE
Flo98
ADAM
Hammerhead
MCSA-PCRAUTODOCK
MCDOCK
ProDOCK
ICM
DockVision
3D-Dock
HEX
GRAMM
PPD
BIGGER
ZDOCK/RDOCK
RosettaDock
Affinity
CombiBUILD
GLIDE
GOLD
HINT
LIGPLOT
SITUS
VEGA
FRED
FlexiDock
8/10/2019 USC NancyDeng MolecularDockingStudies
53/58
(A)
eHiTS Demo
8/10/2019 USC NancyDeng MolecularDockingStudies
54/58
8/10/2019 USC NancyDeng MolecularDockingStudies
55/58
8/10/2019 USC NancyDeng MolecularDockingStudies
56/58
8/10/2019 USC NancyDeng MolecularDockingStudies
57/58
8/10/2019 USC NancyDeng MolecularDockingStudies
58/58